Artificial intelligence has been a popular subject for years but rose to prominent fame with ChatGPT. Even though most are familiar with AI since it’s commonly used by students, many also predicted it would be used in other fields in the future— which is exactly what happened. Some fields where AI is currently being used include robotics, marketing, finance, and healthcare. Healthcare has always been interacting with the latest technology, but the latest developments concerning the use of AI in cancer diagnosis and treatment have taken it to the next level, making it easier for doctors to quickly and precisely diagnose cancer patients and prescribe the most effective medication combinations.
The use of AI software in cancer diagnosis has drastically improved, with the newest software having a 100% accuracy rate for melanoma, one of the most severe cases of skin cancer. The research was displayed at the 2023 European Academy of Dermatology and Venereology Congress and was said to have examined 22,356 patients suspected of skin cancer over 2.5 years. The software was able to recognize 59/59 cases of melanoma, as well as accurately notice 189/190 of all skin cancers and 541/585 of pre-cancerous lesions. This is a remarkable improvement when compared to the first version of the software which was tested in 2021 and detected only 85.9% of melanoma, 83.8% of all skin cancers, and 54.1% of pre-cancerous lesions. Between April 2022 and January 2023, the most recent software version saved over 1,000 in-person consultations in the context of secondary care, granting more time to patients who need rapid attention. Even though the research team's results are extremely encouraging, they made it clear during the conference that the software should not be used as a stand-alone detection tool because it failed to identify skin cancer in one patient, a finding that was later verified by a dermatologist for safety. Additionally, the group emphasized that AI is meant to be a tool to speed up the diagnosing process rather than a replacement for dermatologist consultants.
Cancer diagnosis is not the only extent of AI use in healthcare. In 2022, scientists under the direction of NIHR research professor Udai Banerji developed a prototype AI-based test that could identify the drug combinations that cancer patients are most likely to react to in only one to two days. Mutations that help cancer cells reproduce can be determined via genetic analysis, and some of these mutations can be treated with medications, but knowing the mutations doesn’t mean that doctors always know which mixture of drugs they should use. This is where the AI-based test comes in. The test uses AI to investigate large-scale protein information taken from tumor samples and can predict cancer patients’ response to medication more accurately than what is currently possible. To develop the test, researchers looked at 52 significant proteins' reactions to various drug treatments. They, then, taught machine learning algorithms to identify the critical protein alterations that are predictive of drug responses. The algorithm was then applied to 21 different two-drug combinations in lung cancer cells with various gene defects to predict the sensitivity of cells to drug combinations. 128 of the 252 drug combinations displayed some level of synergy, which means the effect of the two drugs combined was greater than the sum of their individual impacts. The test accurately selected the top five combinations 57% of the time and the top ten 83% of the time. Combinations that have been known to be promising were successfully recognized by the test, along with some new ones like the mix of capivasertib and vemurafenib, which the test found may be useful for non-small cell lung cancer cell lines that do not have EGFR or KRAS mutations. Of course, before the test can start to be used in clinical treatments, it first needs to be experimented on patients who are currently going through cancer treatment. Nevertheless, this AI-powered test is one of the greatest advancements in the fight against cancer and the first prototype test to provide personalized medicine combinations that are likely to be effective.
In conclusion, the use of AI in healthcare, especially in topics concerning cancer, is predicted to increase in the upcoming years if research continues to produce encouraging results like these. However, this does not reject the need for doctors, who should at the very least be examining AI to ensure it does not make any mistakes that could harm patients.
Works Cited
National Institute for Health and Care Research. “AI Test Could Predict Effective Cancer Drug Combinations in Less than Two Days.” Www.nihr.ac.uk, 4 Apr. 2022, www.nihr.ac.uk/news/ai-test-could-predict-effective-cancer-drug-combinations-in-less-than-two-days/30298.
EMJ Dermatol. 2023;11[1]:10-19. DOI/10.33590/emjdermatol/10306603. https://doi.org/10.33590/emjdermatol/10306603.